Is there any non-brutal way like restarting dremio service on the master to forcefully (and gracefully) cancel or removing currently running or queuing jobs from the job list for whatever reason?
For example, a user executed a query that blocking dremio executor i/o which prevented or make the next query running on snail speed.
Or when somehow, the jobs just vanished from the coordinator and stuck on the list because the master doesn’t know whether the job is running, canceled, or encountered an error.
Etc.
In newer versions of Dremio, job cancelation is more consistent. What version of Dremio are you on?
Sorry for the late reply.
We’re currently using v3.3.2 and still deciding which v4.x we will upgrade.